Humphreys’ account of emergence was motivated by the desire to avoid the exclusion argument or a generalized version thereof, whose conclusion is that higher level emergent properties are excluded from affecting lower level prop- erties, since all the causal work is done by the latter (see Kim 1992, 1999, 2006). The exclusion argument has unwelcome consequences for the ontology
of the special sciences. If one thinks of the special science properties (e.g., chemical or biological) as occupying higher levels than do physical properties, then the exclusion argument entails that no event involving a special science property could ever causally influence a physical event. The idea of special science causation is thus threatened. Also, the exclusion argument challenges the idea that special science properties deserve a place in our ontology: if special science properties are causally idle, what is the point of having them in our ontology? The exclusion argument has unwelcome consequences for physics, too. If one thinks of physics itself as stratified (e.g., with high energy physics, solid state physics and thermodynamics occupying different strata), the exclusion argument entails that only the most basic physical properties can be causally efficacious, and – as a result – all other causal claims within contemporary physics are false.
While the exclusion argument denies that the higher level properties that special sciences are concerned with are capable of downward causation, emer- gence seems to require it explicitly. It has been argued that the only way to cause an emergent property to be instantiated is by causing its emergence base property to be instantiated (Kim 1992, p. 136). This is known as the downward causation argument, and it shares with the exclusion argument the assumption that the right way to represent the relation between lower level and higher level properties is supervenience.
In his work, Paul Humphreys challenges both the exclusion argument and the downward causation argument by explicitly denying their common as- sumption, namely that supervenience is the right way to represent the rela- tion between lower level and higher level properties (1997a). He also argues that thinking of higher level emergent properties in terms of supervenience is mistaken. Instead, he links the possibility of emergence with the existence of a fusion operation that operates on i-level properties and outputs i+1-level properties, which have novel causal powers.8
8For the sake of brevity, sometimes I will use “property” instead of “property instance”. It should be noted however that for Humphreys the arguments of the fusion operation are property instances.
The process of fusion is formally represented as follows. Let Pim(xi r)t1 represent an i-level entity, xr, instantiating an i-level property, Pm, at time
t1. Pin(xis)t1 will denote another i-level entity, xs,instantiating another i-level
property, Pn, at time t1. Humphreys introduces the fusion operation symbol- ized by [.∗.], which takes as arguments the two property instances Pim(xi
r)t1 and Pin(xi
s)t1 and fuses them: [Pim(xir)t1∗Pin(xis)t1]. The fusion operation is an
i-level operation, i.e., an operation of the same level as its arguments. The result of the fusion operation is the fused property [Pim∗Pin][(xir)+(xis)](t2) at thei+1-level, which can also be written as [Pil+1][xil+1](t2). The fused property is a unified whole in the sense that its causal effects cannot be represented in terms of the separate causal effects of the original property instances. Also, within the fused property instance [Pim(xi
r)t1∗Pin(xis)t1], the original property instances Pim(xi
r)t1 and Pin(xis)t1 no longer exist as separate entities and they do not have all of theiri-level causal powers available for use at the i+1-level (Humphreys 1997b, p. 10).
Humphreys argues that this particularity of fusion emergence is what en- ables this brand of emergentism to avoid the threats of the exclusion and downward causation arguments. At the time when the fused property instance [Pim(xi
r)t1∗Pin(xis)t1] comes into existence, the original property instances Pim(xir)t1 and Pin(xi
s)t1 go out of existence. Therefore, it is a fortiori the case that they cannot compete as causes with the emergent property instance. On Humphreys account, emergents don’t coexist with their bases, and this feature prevents the exclusion argument from getting off the ground.
Humphreys’ fusion emergence also deals with the downward causation ar- gument. This argument is also committed to the idea that emergent properties supervene on lower level properties. The argument assumes that theonly way to bring about an emergent property instance at timet is by bringing about its subvenience base at time t. But if fusion emergents are not synchronous with their bases, this assumption is unwarranted. There is no reason to suppose that an i+1-level property instance could not directly produce another i+1- level property instance e.g., by directly transforming into it or by transforming another, already existing, i+1-level property instance – in both cases, other
property instances may contribute (1997b, p. 13; 2008, p. 8).
By avoiding the threats to the ontology of the special sciences posed by the exclusion and downward causation arguments, Humphreys’ emergentist account attempts to rescue the autonomy of the special sciences and to de- pict an ontologically antireductionist image of the world in which the subject matters of the various special sciences correspond to irreducible ontological strata.9 For Humphreys, there is a hierarchy of levels of properties L
0,L1, ... Ln... of which at least one distinct level is associated with the subject matter of
each special science, and Lj cannot be reduced to Li for anyi <j (Humphreys
1997a, p. 5).